Abstract – In identification problems, it is necessary to process large amounts of information about transient processes in short time intervals. The increase in identification speed can be achieved due to application of parallel algorithms. Developed symbolical address models for allocation and rerouting of information allow representing parallel computing algorithms in a formalized mathematical form. They possess recursive and regular properties, which facilitates creation of software for solving identification problems. Address models are described in a symbolical form with the use of address lexicographic combinatory configurations formed in conditional addresses. Due to this fact, a connection between the structure of computing algorithms and the architecture of address models for information allocation in algorithms is established. Such approach allows to preserve the parameters of identified object and the parameters of dynamic test modes in address models in a symbolical form. It can have practical importance for imitation modelling of identification algorithms. It allows studying the regular properties of algorithms which in practice can be violated and are a reason of reception of unreliable results. The developed address models can be used for solving difference equation systems of identification and for derivation of their solutions in an analytical form. On the basis of such models, a number of important theoretical results, which are relevant for practical application of identification algorithms, have been obtained.